Founder Effects Persist Despite Adaptive Differentiation: A Field

Experiment with Lizards

Jason J. Kolbe, et al.
Science 335, 1086 (2012);
DOI: 10.1126/science.1209566

If you wish to distribute this article to others, you can order high-quality copies for your
colleagues, clients, or customers by clicking here.
Permission to republish or repurpose articles or portions of articles can be obtained by
following the guidelines here.
The following resources related to this article are available online at
www.sciencemag.org (this information is current as of March 19, 2012 ):
Updated information and services, including high-resolution figures, can be found in the online
version of this article at:
http://www.sciencemag.org/content/335/6072/1086.full.html
Supporting Online Material can be found at:
http://www.sciencemag.org/content/suppl/2012/02/02/science.1209566.DC1.html
This article cites 30 articles, 6 of which can be accessed free:
http://www.sciencemag.org/content/335/6072/1086.full.html#ref-list-1
This article appears in the following subject collections:
Evolution
http://www.sciencemag.org/cgi/collection/evolution

Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by the
American Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. Copyright
2012 by the American Association for the Advancement of Science; all rights reserved. The title Science is a
registered trademark of AAAS.

Downloaded from www.sciencemag.org on March 19, 2012

This copy is for your personal, non-commercial use only.

REPORTS
Acknowledgments: We thank T. Giesecke and M. Lascoux for
helpful comments; J. Chen, T. Geburek, and C. Sperisen for
providing modern samples; and L. Yuan for initial spruce
genotyping. We also thank T. Vorren and K.-D. Vorren for
providing Andya core samples with lithostratigraphy and
radiocarbon dates. This study was supported by the Swedish
Research Council (grant 2007-4490 to L.P.) and the Carl
Tryggers Foundation (grant 08:303 to L.P.); KAKENHI
(22658046); a Royal SocietyWolfson Merit Award to K.D.B.;
the Danish National Research Foundation; the Norwegian
Forest and Landscape Institute, University of Troms; and the
Troms University Museum, the Roald Amundsen Centre for
Arctic Research, and the European Commission under the Sixth
Framework 378 Programme (EcoChange project, contract no.
FP6-036866). L.P., T.J., M.M.T., and E.E. should be considered
joint first authors. I.G.A. and E.W. should be considered joint
senior authors. The authors declare no competing financial
interests. P.T. is coinventor of patents related to the g/h
primers and the use of the P6 loop of the chloroplast trnL

Founder Effects Persist Despite

Adaptive Differentiation: A Field
Experiment with Lizards
Jason J. Kolbe,1* Manuel Leal,2 Thomas W. Schoener,3 David A. Spiller,3 Jonathan B. Losos1
The extent to which random processes such as founder events contribute to evolutionary
divergence is a long-standing controversy in evolutionary biology. To determine the respective
contributions of founder effects and natural selection, we conducted an experiment in which
brown anole ( Anolis sagrei ) lizard populations were established on seven small islands in the
Bahamas, from male-female pairs randomly drawn from the same large-island source. These
founding events generated significant among-island genetic and morphological differences
that persisted throughout the course of the experiment despite all populations adapting in the
predicted directionshorter hindlimbsin response to the narrower vegetation on the small
islands. Thus, using a replicated experiment in nature, we showed that both founder effects and
natural selection jointly determine trait values in these populations.

sland populations are renowned for their extent of divergence from each other and from
mainland source populations (1, 2). Mayr
(3, 4) argued that these differences are often
triggered by random sampling when island populations are founded by a few colonizing individuals. The resultant founder effectschanges in
the genetic and phenotypic composition of a population due to founding by a small number of
individualshave been proposed as an important
cause of evolutionary divergence and even speciation for the past half-century (36). However,
an alternative explanation is that island environments differ from each other and from the source
locality, and these ecological differences result in
divergent natural selection (79). The evolutionary significance of founder effects also has been
1
Department of Organismic and Evolutionary Biology and Museum of Comparative Zoology, Harvard University, 26 Oxford
Street, Cambridge, MA 02138, USA. 2Department of Biology,
Duke University, Durham, NC 27708, USA. 3Section of Evolution and Ecology and Center for Population Biology, One Shields
Avenue, University of California Davis, Davis, CA 95616, USA.

*To whom correspondence should be addressed. E-mail:

[email protected]
Present address: Department of Biological Sciences, University of Rhode Island, Kingston, RI 02881, USA.

1086

questioned because their imprint may be shortlived if populations perish because of a lack of
genetic variation or because of demographic stochasticity, or if natural selection overwhelms their
effects (10, 11). Data from nature are lacking because founder events are rarely observed [but see
(12)], and thus their effects must be inferred post
hoc; yet laboratory studies indicate that even in
the presence of natural selection, genetic drift induced by founder effects or population bottlenecks
can contribute to patterns of phenotypic divergence [e.g., (13, 14)]. We report an experimental
study of founder effects in a natural system of a
Caribbean lizard, demonstrating that morphological divergence caused by the founder effect
persists even as populations adapt to their new
environments.
Some have argued that marked phenotypic
change from a single founder event is very unlikely (11) and that genetic drift in general plays little
part in morphological evolution (15). Moreover,
others note that evidence from recent introductions suggests that adaptation is a more common
cause of trait evolution than drift (1618). Yet
most concede that documenting these random
sampling events in nature is exceedingly difficult
(11, 15, 1921). Without knowledge of founder

2 MARCH 2012

VOL 335

SCIENCE

(UAA) intron for plant identification using degraded template

DNA. These patents only restrict commercial applications
and have no impact on the use of this locus by academic
researchers. The trnL DNA sequences can be found at the
European Nucleotide Archive with the study accession no.
ERP001153. The mtDNA sequences can be found at the
European Molecular Biology Laboratory database with
the accession no. HE652882-HE653002.

Supporting Online Material

www.sciencemag.org/cgi/content/full/335/6072/1083/DC1
Materials and Methods
SOM Text
Figs. S1 to S5
Tables S1 to S5
References (4092)
1 November 2011; accepted 26 January 2012
10.1126/science.1216043

attributes and repeated sampling of colonized islands, the relative contributions of founder effects, population bottlenecks, natural selection, and
gene flow to evolutionary divergence are impossible to disentangle (1921).
We capitalized on the extensive knowledge
of how Anolis lizards adapt to their environment,
combined with the opportunity to use small islands in the Bahamas, recently cleared of lizards
by a hurricane, as experimental units in a natural
setting. Our focus was on limb length and the
extent to which lizard populations would adapt to
the novel environments on these small and scrubby islands. Extensive comparative and experimental research supports an adaptive explanation for
the positive relationship between hindlimb length
and perch diameter (i.e., the width of the cross section of the substrate that a lizard perches on, such
as a branch or trunk) that is found among populations of anoles. Functional studies show a
clear biomechanical basis: Lizards with relatively
longer hindlimbs run faster on broad substrates,
whereas lizards with shorter limbs for a given
body size move more adeptly on narrow surfaces.
Such performance probably aids in capturing prey,
defending territories, and escaping from predators. Indeed, field studies in this system show that
hindlimb length is under selection, favoring longer
hindlimbs to run faster when exposed to terrestrial predators and shorter hindlimbs after lizards
become arboreal and shift to narrower perches
[reviewed in (22)].
To assess the relative importance of founder
effects versus natural selection, we introduced lizards to replicate islands to which we predicted
they were not well adapted because of differences
in structural habitat from their source, thereby simulating founding events and altering the selective
regime for limb length. Specifically, founding propagules were established on small experimental
islands, all of which are sparsely vegetated and
covered primarily with short, narrow-diameter vegetation as compared to the more forested habitat
of the lizards source area on a nearby larger island (Fig. 1 and figs. S1 and S2). Our prediction
is that if natural selection is the dominant force,
then we would expect all populations to evolve
shorter hindlimbs as they adapt to using narrower

www.sciencemag.org

Downloaded from www.sciencemag.org on March 19, 2012

27. K. Andersen et al., Mol. Ecol. 10.1111/j.1365-.2011.05261.x

(2011).
28. A. Paus, G. Velle, J. Berge, Quat. Sci. Rev. 30, 1780 (2011).
29. T. Alm, Boreas 22, 171 (1993).
30. T. O. Vorren, K.-D. Vorren, T. Alm, S. Gulliksen, R. Lvlie,
Boreas 17, 41 (1988).
31. K. D. Vorren, Boreas 7, 19 (1978).
32. T. Alm, H. H. Birks, Nord. J. Bot. 11, 465 (1991).
33. P. Taberlet et al., Nucleic Acids Res. 35, e14 (2007).
34. T. Preuten et al., Plant J. 64, 948 (2010).
35. U. Rauwolf, H. Golczyk, S. Greiner, R. G. Herrmann,
Mol. Genet. Genomics 283, 35 (2010).
36. J. Haile et al., Mol. Biol. Evol. 24, 982 (2007).
37. L. Parducci, Y. Suyama, M. Lascoux, K. D. Bennett,
Mol. Ecol. 14, 2873 (2005).
38. T. O. Vorren, L. Plassen, Boreas 31, 97 (2002).
39. J. Mangerud, in Quaternary GlaciationsExtent and
Chronology, Part 1Europe (Elsevier, Amsterdam,
2004), pp. 271294.

REPORTS
Fig. 4, inset). As with the genetic data, hindlimb
values for island populations were dispersed around
the mean value of the source population on Iron
Cay (fig. S3). Furthermore, there was no relationship between perch diameter and relative hindlimb
length among islands in 2006 (r2 = 0.15, P = 0.39),
supporting the interpretation that differences in
hindlimb length among islands resulted from the
founder event rather than adaptive divergence.

Fig. 1. Vegetation differences between the source population and those of the experimental founder
islands. The schematic cartoon shows the change in the vegetation profile for lizards introduced from the
source population on the more forested Iron Cay to the sparsely vegetated experimental founder islands.
The vegetation illustrated is scaled to the mean values of vegetation height for Iron Cay and the pooled
experimental founder islands, respectively. Bar graphs show the difference (mean T SE) in available vegetation height and diameter from systematic transects on each island. We predicted that the change in
vegetation profile would result in lizards using narrower perches on experimental founder islands, which
would alter the selective regime to favor shorter hindlimbs.

Downloaded from www.sciencemag.org on March 19, 2012

after the founding event; only 1 locus out of 42

on the seven islands (six loci per island) showed a
net decrease in the number of alleles from founding in 2005 to 2009. These results support the
persistence of the initial founder effect without
subsequent genetic drift through time.
A founder effect also was evident for hindlimb
length, with experimental founder islands differing significantly in 2006 (P = 0.0015; table S3 and

Fig. 2. Population size

45
estimates for experimenExperimental
40
Founder Islands:
tal founder islands. Surveys
N1
were conducted in May of
35
each year from 2005 to
N15
30
2009 [see (29) for details].
N2
Population size showed a
25
N3
positive relationship with
N4
20
vegetated area of each
X10
island in 2006 (r 2 = 0.53,
15
X3*
2
P = 0.06) and 2007 (r =
10
0.74, P = 0.01), suggesting an effect of habitat
5
area on population growth
0
over the first 2 years. Pop2005
2006
2007
2008
2009
ulation size estimates from
Year
November 2006, which included hatchlings produced
that year, ranged from 21.4 to 60.2 (not shown in this figure). In all but one case, the November 2006 size
estimates greatly exceeded the population size in May 2007, suggesting high mortality rates. Rapid
population expansion such as found here may produce conditions favorable to evolution by natural
selection (16, 30) and may also curtail the loss of genetic variation after the initial founder effect.

www.sciencemag.org

Population size

substrates; moreover, vegetation differences among

experimental founder islands would be expected
to produce a relationship between hindlimb length
and perch diameter [reviewed in (22)]. Conversely, if founder effects are dominant, then we would
expect no general trend in limb length evolution,
with some populations increasing in limb length
and others decreasing with respect to the source
population, and limb variation being unrelated to
vegetation differences among islands.
In May 2005, we randomly selected one male
and one female brown anole from lizards collected
on a nearby larger island (Iron Cay, Bahamas,
vegetated area >150,000 m2) to found populations on seven small islands (vegetated area
35 to 175 m2). The storm surge from Hurricane
Frances in September 2004 submerged these small
islands near Great Abaco, extirpating their lizard
populations (23). We took morphological measurements and tissue samples for later genetic
analysis from these founders. Over the next 4 years,
we characterized the environment and repeatedly
sampled lizards from the source population (Iron
Cay), from the seven experimental founder islands, and from 12 nearby reference islands (vegetated area 74 to 324 m2) having A. sagrei and a
composition of spider, insect, and plant species
similar to that of the experimental islands (24).
All lizard populations increased in size for the
first 2 years (averaging a 13-fold increase) and
fluctuated in size thereafter (Fig. 2).
A founder effect was immediately apparent,
conforming to theoretical expectations and previous empirical studies of genetic variation (2527).
In 2006, 1 year after introduction, multilocus genotypes from six microsatellite loci revealed an average decrease of 46% in allelic diversity and
23% in heterozygosity on experimental founder
islands as compared to their source. Furthermore,
allele frequencies differed significantly among
all pairs of populations, including the source on
Iron Cay (Stouffers z method of combining
probabilities from exact G tests of six microsatellite loci; all P < 0.05; table S1). Founder populations were dispersed around the mean value
of the source population (Iron Cay) and varied
fivefold in their amount of divergence from it;
moreover, among-islands genotype frequencies
in subsequent years were correlated with the genotypes of the two founding individuals [correlation
coefficient (r) = 0.80 to 0.97 on principal coordinates
(PCo) axes 1 to 3 for 2006, 2007, and 2009, all
P < 0.05; Fig. 3]. Islands differed significantly
from each other in genotype frequencies (analyses of variance for PCo 1 to 3 separately, F6,248 =
16.5 to 41.7, all P < 0.0001), whereas differences
both across years and among islands across years
were not significant. In all but one case (the island N3Iron Cay comparison), pairwise fixation
index (FST) values indicated significant genetic
differentiation between islands (table S2), although FST values should be interpreted cautiously because of the likelihood of nonequilibrium
conditions in these recently established populations. Moreover, genetic diversity was sustained

SCIENCE

VOL 335

2 MARCH 2012

1087

REPORTS

0.03

Relative hindlimb length

Change in relative hindlimb length per year

0.03

0.02

0.01

0.02
0.01

N1

N15
N3
N4

-0.01

X3*
N2

-0.02

X10

-0.03
2006

2007

2009

Year
-0.01

-0.02

-0.03

Reference
Islands

Experimental
Founder Islands

Fig. 4. Change in hindlimb length on reference and experimental founder islands. Change in mean
relative hindlimb length [calculated as the residuals of the regression of loghindlimb length on log-body
length (i.e., snout-vent length), separately for each sex] per year for populations on experimental founder
and reference islands, including the source population on Iron Cay (open circle). The inset in the upper
right shows how the mean (TSE) relative hindlimb length decreases on the experimental founder islands
in each year. Islands N1, N3, and X3* are shown with dotted lines for clarity. Immigration does not alter
the results for hindlimb differentiation, which are virtually identical when putative immigrants are removed from analyses. Although the source population (Iron Cay) decreased in hindlimb length, all lizards
on experimental founder islands exhibited even greater decreases than those on Iron Cay and on all other
reference islands, with the mean rate of hindlimb change on reference islands not differing from zero.

1088

2 MARCH 2012

VOL 335

SCIENCE

We used systematic transects to quantify the

available structural habitat (i.e., vegetation height
and diameter). The vegetation profile of the source
population on Iron Cay averaged substantially
taller and broader vegetation (mean vegetation
height = 384 cm and mean vegetation diameter =
11.1 cm) as compared to the experimental founder islands, which have much shorter and narrower
vegetation (mean vegetation height ranged from
62 to 107 cm and mean vegetation diameter ranged
from 1.6 to 5.4 cm) (Fig. 1 and figs. S1 and S2).
In turn, the mean perch diameter used by lizards
on all experimental founder islands (3.2 cm) was
smaller than for the source population (4.6 cm).
As predicted by the adaptive relationship between hindlimb length and substrate diameter (22),
relative hindlimb length decreased greatly over
the course of 3 years on all seven experimental
founder islands (P < 0.0001; table S4 and Fig. 4,
inset), with a decrease of 6.5% for males and
4.0% for females (calculated at the median body
size for each sex; table S5 and fig. S4). The magnitude of this decrease in hindlimb length was so
great that mean values in 2006 are nearly nonoverlapping with 2009 values (Fig. 4, inset). We
monitored nearby reference islands for the same
length of time, and the decrease on experimental
founder islands was five times greater than on the
reference islands [mixed effects analysis of covariance (ANCOVA) test for an island type (reference
versus experimental founder islands)byyear
interaction: P < 0.0001; table S6 and Fig. 4]. The
rates of decrease on all seven founder islands
were greater than the rates for all reference islands, including the source population on Iron
Cay (Fig. 4), and the mean rate for the reference

www.sciencemag.org

Downloaded from www.sciencemag.org on March 19, 2012

PCo 2 (21%)

6
Fig. 3. Multilocus genotypic variation for the
source population and
those of the experiment4
al founder islands. We
conducted a PCo analysis
of six microsatellite loci
2
(mean T SE of PCo values). Numbers within each
symbol correspond to different sampling years: I,
0
2005 (i.e., the founding
Iron Cay
pair on each experimentN1
al founder island and
-2
N2
these same founder indiN3
viduals pooled for the
N4
source population estimate
on Iron Cay); II, 2006 (no
N15
-4
Iron Cay estimate); III,
X3*
2007; and IV, 2009. The
X10
percentage of variation
-6
explained for each PCo
-12
-10
-8
-6
-4
-2
0
2
4
6
axis is in parentheses.
Stored sperm use was
PCo 1 (29%)
detected in 6% of offspring through 2006, resulting in offspring fathered by a male other than the one introduced onto the island. However, all other individuals sampled during this
time had multilocus genotypes consistent with being progeny of the founding pair (or founding female in the case of stored sperm use), and not until 2007 did
we detect unambiguous immigrants.

islands did not differ significantly from zero (t =

1.55, df = 12, P = 0.15), whereas the mean rate
for the experimental founder islands did (t =
11.53, df = 6, P < 0.0001).
Despite this substantial adaptive response,
the mark of the founder event persisted, as illustrated by the significant difference among islands (ANCOVA: P < 0.0001; table S4) and the
minimal change in the order of mean hindlimb
length values among experimental founder islands
from 2006 to 2009 (year-by-island interaction:
P = 0.70; table S4 and Fig. 4, inset). Furthermore,
multiple regression confirms the persistence of
the founder effect by showing that mean relative
hindlimb lengths among islands in 2006 had a
positive relationship with values in 2009 (P =
0.046), and it demonstrates that no relationship
existed between hindlimb lengths in 2009 and the
diameter of either available vegetation (P = 0.17)
or perches (P = 0.86), indicating that adaptation
to differing environments cannot explain variation among islands in hindlimb length.
Hindlimb growth can be affected by environment during ontogeny in A. sagrei (28), but such
phenotypic plasticity is unlikely to explain the
large decreases in hindlimb length detected on
the experimental founder islands from 2006 to
2009. If plasticity were important, we would have
expected to see an immediate difference in hindlimb length between the founders and their firstgeneration descendants due to the latter developing
in the new, narrow-vegetation environment, but
the range in population means in 2006 overlapped
completely with that of the male founders. Moreover, perch diameter did not change over time on
experimental founder islands (year: P = 0.10, and
year-by-island interaction: P = 0.13), even as hindlimb length progressively decreased, showing that
the change in hindlimb length cannot be attributed
to a change in a possible stimulus for plasticity.
Finally, the magnitude of hindlimb divergence is
three to five times greater than that produced by
plasticity in A. sagrei raised on only narrow or broad
surfaces in laboratory experiments (28) (fig. S5),
even though the difference in perch diameter
(1 versus 9 cm) imposed in the laboratory experiment was much greater than the differences in
mean perch diameter observed between the source
population and those of the experimental founder
islands (4.6 cm and 2.6 to 4.5 cm for Iron Cay
and the experimental founder islands, respectively).
Founder events generated among-island genetic and phenotypic differences that were similar
in magnitude to those observed among populations on nearby islands (our reference islands),
as well as among natural populations occurring
throughout the Bahamas (tables S1 to S3, Fig. 3,
and figs. S3 and S5). The strong genetic differentiation among founder islands, illustrated by allelic differentiation and FST values of similar
magnitude to those seen among nearby reference
islands, remained stable over the 4 years of the
study (Mantels r = 0.86 to 0.94 for correlations
among pairwise FST matrices in 2006, 2007, and
2009, all P < 0.01; tables S1 and S2). The strong

imprint of the founder effect and the stability of

among-population genetic differentiation occurred
despite evidence for ongoing immigration. Assignment tests and pedigree analyses revealed
that 11% of individuals sampled on experimental
founder islands in 2007 and 2009 had multilocus
genotypes consistent with being first-generation
immigrants. In terms of phenotypic differentiation,
the level of hindlimb divergence observed among
experimental founder islands was only slightly
less than that seen both among nearby reference
islands and among natural populations from across
the Bahamas (fig. S5). These results did not change
when putative immigrants were removed from
analyses. That the magnitude of the genetic and
phenotypic differentiation produced by founder
events rivals that seen among populations on natural islands indicates that founder effects may, as
some have suggested, be an important cause of
variation among islands (3, 4).
Our results indicate that founder effects persist, even in the face of substantial adaptive differentiation. The impacts of founder events on
neutral genetic variation have long been recognized and are clearly evident here (25, 27) (tables
S1 and S2 and Fig. 3), but the impact of founder
effects on phenotypic divergence is still debated.
In support of the selectionist school, we show that
differentiation from the source population is mostly
the result of adaptation to the new environmental
milieu on the experimental islands (79, 1618, 22).
However, the imprint of the founder effect remains
apparent, even as this adaptive divergence has
occurred; indeed, variation among experimental
founder islands at the present time is better explained by initial phenotypes than by current environmental conditions (Fig. 4, inset). What remains
to be seen is whetheror for how longthese
founder effects will persist in the face of natural
selection. One might expect that just as selection
has driven a decrease in hindlimb length from that
of the source population in response to the narrower
vegetation on the founder islands, it will ultimately also drive smaller-scale differentiation among
islands as their lizard populations similarly adapt
to the much smaller differences in perch diameter
and erase the signal of the founder event. This prediction is not a foregone conclusion: Not only are
habitat differences among experimental founder
islands small, but the ultimate outcome is dependent on the interaction of changing demographic,
environmental, and genetic factors.
We here show that founder effects can play an
important role in the divergence of island populations, even as adaptation occurs. By measuring the phenotypic values of founders and their
descendents, we have distinguished between the
influences of the founder effect and of adaptation;
thus, we provide clear evidence of a rarely characterized dimension of evolutionary divergence
among populations.
References and Notes
1. P. R. Grant, Ed., Evolution on Islands (Oxford Univ. Press,
Oxford, 1998).

www.sciencemag.org

SCIENCE

VOL 335

2. R. G. Gillespie, D. A. Clague, Eds., Encyclopedia of

Islands (Univ. of California Press, Berkeley, CA, 2009).
3. E. Mayr, in Evolution as a Process, J. Huxley, A. C. Hardy,
E. B. Ford, Eds. (Allen & Unwin, London, 1954),
pp. 157180.
4. E. Mayr, Animal Species and Evolution (Harvard Univ.
Press, Cambridge, MA, 1963).
5. H. L. Carson, Stadler Genet. Symp. 3, 51 (1971).
6. A. R. Templeton, Genetics 94, 1011 (1980).
7. J. A. Endler, Natural Selection in the Wild (Princeton
Univ. Press, Princeton, NJ, 1986).
8. N. H. Barton, in Evolution on Islands, P. R. Grant, Ed.
(Oxford Univ. Press, Oxford, 1998), pp. 102123.
9. D. Schluter, The Ecology of Adaptive Radiation (Oxford
Univ. Press, Oxford, 2000).
10. T. Price, Speciation in Birds (Roberts & Co., Greenwood
Village, CO, 2007).
11. N. H. Barton, B. Charlesworth, Annu. Rev. Ecol. Syst. 15,
133 (1984).
12. P. R. Grant, B. R. Grant, K. Petren, Genetica 112-113,
359 (2001).
13. M. Travisano, J. A. Mongold, A. F. Bennett, R. E. Lenski,
Science 267, 87 (1995).
14. P. Simes et al., Evolution 62, 1817 (2008).
15. J. A. Coyne, N. H. Barton, M. Turelli, Evolution 51,
643 (1997).
16. D. N. Reznick, C. K. Ghalambor, Genetica 112-113,
183 (2001).
17. S. M. Clegg et al., Evolution 56, 2090 (2002).
18. M. T. Kinnison, N. G. Hairston Jr., Funct. Ecol. 21, 444
(2007).
19. C. W. Kilpatrick, in Mammalian Population Genetics,
M. H. Smith, J. Joule, Eds. (Univ. of Georgia Press,
Athens, GA, 1981), pp. 2859.
20. R. J. Berry, in Evolution on Islands, P. R. Grant,
Ed. (Oxford Univ. Press, Oxford, 1998), pp. 3550.
21. P. R. Grant, Proc. Natl. Acad. Sci. U.S.A. 99, 7818
(2002).
22. J. B. Losos, Lizards in an Evolutionary Tree: Ecology and
Adaptive Radiation of Anoles (Univ. of California Press,
Berkeley, CA, 2009).
23. The patterns of lizard extinction were similar to those
documented for Hurricane Floyd in (31).
24. D. A. Spiller, T. W. Schoener, Ecology 88, 37 (2007).
25. M. Nei, T. Maruyama, R. Chakraborty, Evolution 29,
1 (1975).
26. R. Chakraborty, M. Nei, Evolution 31, 347 (1977).
27. C. L. Tarr, S. Conant, R. C. Fleischer, Mol. Ecol. 7, 719 (1998).
28. J. B. Losos et al., Evolution 54, 301 (2000).
29. T. W. Schoener, D. A. Spiller, J. B. Losos, Ecol. Monogr.
72, 383 (2002).
30. D. Reznick, H. Rodd, L. Nunney, in Evolutionary
Conservation Biology, R. Ferrire, U. Dieckmann,
D. Couvet, Eds. (Cambridge Univ. Press, Cambridge,
2004), pp. 101118.
31. T. W. Schoener, D. A. Spiller, J. B. Losos, Science 294,
1525 (2001).

Downloaded from www.sciencemag.org on March 19, 2012

REPORTS

Acknowledgments: We thank B. Pinder, J. Piovia-Scott,

D. Steinberg, Y. Stuart, C. Thornber, and C. White for
assistance; the National Geographic Society (grant 8704-09),
NSF (grants DEB-0444763 and DEB-0949415), and Duke
University for funding; and the Bahamas Ministry of
Agriculture and the Bahamas Environment, Science &
Technology Commission of the Ministry of the Environment for
permission to conduct this research. This material was based
on work supported by NSF while one of the authors (D.A.S.)
was working at NSF. Any opinions, findings, and conclusions or
recommendations expressed in this material are those of the
authors and do not necessarily reflect the views of NSF.

Supporting Online Material

www.sciencemag.org/cgi/content/full/science.1209566/DC1
Materials and Methods
Figs. S1 to S5
Tables S1 to S6
References (3245)
9 June 2011; accepted 17 November 2011
10.1126/science.1209566

2 MARCH 2012

1089